Applying Causal Forest of Machine Learning to Analyze the Heterogeneous Impact of China's Dual-Carbon Policy on Green Management Innovation

With the relentless advancement of computer technology and the proliferation of machine learning algorithms, the causal forest method has emerged as a pioneering tool in the realm of exploring and unraveling complex causal relationships. The causal forest, a sophisticated extension of random forests, leverages ensemble learning techniques to estimate heterogeneous treatment effects with unprecedented accuracy. In the context of China's ambitious “Dual-carbon Goals,” this study harnesses the power of the causal forest method to delve into the nuanced causal effects of Dual-carbon policies on corporate green management innovation and its underlying heterogeneity. To conduct this comprehensive research, we have meticulously constructed a dataset encompassing Chinese listed companies from 2008 to 2023. By capitalizing on the temporal variation surrounding the implementation of these policies, our analysis systematically dissects the heterogeneous response mechanisms exhibited by firms with diverse characteristics. Empirical results reveal that the Dual-carbon Policy significantly promotes corporate investment and practices in green management innovation. However, the policy effect exhibits notable heterogeneity across firm types: large enterprises, endowed with resource and technological advantages, demonstrate higher levels of green management innovation, while small and medium-sized enterprises (SMEs) face greater cost pressures, resulting in weaker innovation outcomes. Furthermore, significant disparities in policy responsiveness are observed across industries, with high-pollution sectors experiencing more pronounced impacts from the Dual-carbon Policy. Based on these profound insights, this study puts forth policy optimization recommendations that are precisely tailored to diverse enterprise types, with the aim of accelerating the attainment of the “Dual-carbon” targets and facilitating a more profound green transformation. The conclusions of this research not only furnish crucial empirical evidence for governments to craft more precise and effective policies but also underscore the immense potential of computer technology and advanced machine learning methods, such as the causal forest, in propelling sustainable development and enhancing environmental policy-making.